RAJIV KHANNA, 1 * SCOTT R. BURROWS, 1 ANNE NEISIG, 2 JACQUES NEEFJES, 2 DENIS J. MOSS, 1 AND SHARON L. SILINS 1

Transcription

1 JOURNAL OF VIROLOGY, Oct. 1997, p Vol. 71, No X/97/$ Copyright 1997, American Society for Microbiology Hierarchy of Epstein-Barr Virus-Specific Cytotoxic T-Cell Responses in Individuals Carrying Different Subtypes of an HLA Allele: Implications for Epitope-Based Antiviral Vaccines RAJIV KHANNA, 1 * SCOTT R. BURROWS, 1 ANNE NEISIG, 2 JACQUES NEEFJES, 2 DENIS J. MOSS, 1 AND SHARON L. SILINS 1 EBV Unit, Queensland Institute of Medical Research, The Bancroft Centre, Brisbane, Australia 4029, 1 and Division of Cellular Biochemistry, The Netherlands Cancer Institute, 1066 CX Amsterdam, The Netherlands 2 Received 5 May 1997/Accepted 26 June 1997 Major histocompatibility complex class I-restricted Epstein-Barr virus (EBV)-specific cytotoxic T lymphocytes (CTLs) in healthy virus carriers constitute a primary effector arm of the immune system in controlling the proliferation of virus-infected B cells in vivo. These CTLs generally recognize target epitopes included within the latent antigens of the virus. For example, CTLs from HLA B44 healthy virus carriers often recognize peptide EENLLDVFRM from EBV nuclear antigen 6. However, the strength of this response directly correlates with the HLA B44 subtype expressed by the individual donor. Indeed, HLA B*4405 virus carriers consistently show a very high frequency of CTL precursors for the EENLLDVFRM epitope, while a much weaker response is seen in HLA B*4403 and HLA B*4402 individuals. This disparity is not due to an intrinsic difference in the CTLs generated by individuals carrying different subtypes of HLA B44. In fact, virus-specific CTLs recognize EENLLDVFRM peptide-sensitized HLA B*4405 target cells more efficiently than B*4402 or B*4403 target cells irrespective of the HLA B44 subtype expressed by the donors from whom these effectors were isolated. This effect is evident whether the CTL epitope is endogenously processed or exogenously presented. In addition, a comparison of the intracellular transport kinetics of different B44 subtypes revealed that the B*4405 allele is rapidly assembled and arrives in the trans-golgi compartment at a faster rate than B*4402 or B*4403. Based on these results, we propose that HLA class I alleles that are capable of binding peptides more efficiently from the intracellular pool, and are rapidly assembled and transported, may confer a protective advantage against viral infection. * Corresponding author. Mailing address: Queensland Institute of Medical Research, Bancroft Centre, 300 Herston Rd., Brisbane, Australia Phone: Fax: CD8 cytotoxic T lymphocytes (CTLs) recognize peptide epitopes, mainly derived from cytoplasmic and nuclear proteins in association with major histocompatibility complex (MHC) class I molecules on the surface of appropriate antigen-presenting cells (32). Although synthetic peptides can directly prime target cells for lysis by CTLs, recognition of a native antigen requires intracellular processing, which involves cleavage of the antigen into short peptides and their transport into the endoplasmic reticulum (ER), where they associate with MHC class I (1). The binding of these peptides to newly synthesized MHC class I molecules is important for normal assembly and stability of the MHC and for efficient CTL recognition. The number of peptides within an antigenic protein that can bind to a particular class I molecule is usually considerably larger than the number of epitopes recognized in a CTL response to that antigen, suggesting the existence of a hierarchy with which peptides are presented for immune recognition (8, 26, 28). As a consequence, the CTL response is frequently dominated by a limited number of epitopes. The class I binding affinities described so far for immunodominant epitopes are generally very high (10, 16, 30). While it is clear that many immunodominant epitopes exhibit a strong binding affinity, the converse does not always apply; i.e., not all strongly binding peptides are immunodominant epitopes (34). A number of possible mechanisms have been proposed to explain this selective immunodominance: the influence of proteasome-mediated proteolysis of the native antigen (17, 25), peptide affinity for MHC molecules (5, 33), MHC-peptide complex stability (11), and the T-cell repertoire (6). Furthermore, the availability of peptides for the assembly of MHCpeptide complexes in the ER, which has previously been shown to be restricted by the activity of the transporters associated with antigen processing, is also an important factor influencing the level of peptide presentation (22, 24, 29). It is not clear, however, to what extent intracellular folding and transport of MHC molecules influence T-cell repertoire selection and immunodominance in a class I-restricted CTL response. In this study, we have investigated the issue of CTL immunodominance in the context of Epstein-Barr virus (EBV) in which CTLs play a pivotal role in restricting the proliferation of EBV-infected B cells (13). Earlier studies of healthy virus carriers have shown that this CTL response is mainly directed to EBV nuclear antigens (EBNA2 to EBNA6) and latent membrane proteins (12, 20). Multiple peptide epitopes, restricted through different MHC class I antigens, have been identified within these latent antigens (13). We have previously proposed that the level and specificity of CTL responses in healthy virus carriers is strongly influenced by the host MHC genotype (13). The CTL response to the HLA B44-restricted epitope EENLLDVFRM from EBNA6 in individuals carrying different subtypes of this allele is a classic example of this phenomenon. Although the peptide epitope EENLLDVFRM can be presented by three different subtypes of HLA B44 (B*4402, B*4403, and B*4405), individuals carrying these subtypes display a discrete hierarchy in their responses to this epitope. HLA B*4405 virus carriers consistently show a strong response to EENLLDVFRM, while a weak response is 7429

2 7430 KHANNA ET AL. J. VIROL. seen in individuals carrying the HLA B*4403 and HLA B*4402 alleles. We have now investigated the molecular basis of this hierarchy. The data presented in this study show that a strong CTL response to EENLLDVFRM in HLA B*4405 virus carriers correlates with more efficient presentation of the peptide and intracellular transport kinetics of this allele than of HLA B*4402 and HLA B*4403. MATERIALS AND METHODS Establishment and maintenance of cell lines. EBV-transformed lymphoblastoid cell lines (LCLs) from seropositive donors DM (HLA A24, A29, B47, B*4405), IT (HLA A1, A3 B7 B*4402), CM (HLA A11, A24, B7, B*4405), PM (HLA A11, A29, B7 B*4403), CS (HLA A3, A23, B35, B*4403), SE (HLA A2, A29, B60, B*4403), BK (HLA A1, A2, B8, B*4403), DY (HLA A11, A32, B35, B*4402), CF (HLA A1, A2, B8, B*4402), and KE (HLA A32, A26, B49, B*4402) were established by exogenous virus transformation of peripheral B cells, using EBV derived from the B95.8 cell line (19). All LCLs were routinely maintained in RPMI 1640 containing 2 mm glutamine, 100 IU of penicillin per ml, 100 g of streptomycin per ml, and 10 to 20% fetal calf serum (FCS) (growth medium). To generate phytohemagglutinin (PHA) blasts, unfractionated mononuclear (UM) cells were stimulated with PHA (Commonwealth Serum Laboratories, Melbourne, Australia), and after 3 days, growth medium containing MLA-144 supernatant and recombinant interleukin-2 (ril-2) was added (referred to as T-cell growth medium) (4). PHA blasts were propagated with biweekly replacement of ril-2 and MLA-144 supernatant (no further PHA added) for up to 6 weeks. Source and generation of EBV-specific CTL clones. HLA B44-restricted EBVspecific CTL clones recognizing the CTL epitope, EENLLDVFRM, were established from healthy seropositive donors DM (B*4405 ), CS (B*4403 ), SE (B*4403 ), IT (B*4402 ), KE (B*4402 ), and CF (B*4402 ) as described earlier (2, 19). Briefly, peripheral blood lymphocytes from these donors were stimulated in 2 ml of growth medium with autologous -irradiated (8,000 rads) B95.8-transformed LCLs (responder/stimulator ratio of 50:1). After 3 days, cells were dispersed and seeded in 0.35% agarose (SeaPlaque; FMC Corp., Rockland, Maine) containing T-cell growth medium. Colonies were harvested after a further 3 to 4 days and amplified in culture with biweekly restimulation with -irradiated (8,000 rads) autologous LCLs in T-cell growth medium. The EBV and antigen specificity of each CTL clone was confirmed using autologous LCLs and recombinant vaccinia virus constructs encoding individual EBV antigens (12, 14, 15). The peptide specificity of each CTL clone was also determined by using synthetic peptides (4, 18). Generation of B*4402/B*4405 and B*4403/B*4405 chimeras. To generate B*4402/B4405 and B*4403/B*4405 chimeras, LCLs from donors CF (B*4402 ) and BK (B*4403 ) were fused with LCLs from donor CM (B*4405 ), using polyethylene glycol (molecular weight, 2,000). These fused cells were cultured for 72 h at 37 C, and B*4402/B*4405 and B*4403/B*4405 chimeras were separated with a FACS Vantage (Becton Dickinson and Co., Mountain View, Calif.), using HLA-specific monoclonal antibodies. It is important to mention here that donors CF and BK are positive for HLA B8, while donor CM is positive for HLA B7. For fluorescence-activated cell sorting, cells were initially incubated with fluorescein isothiocyanate-labeled HLA B7-specific monoclonal antibody (Behring Diagnostics, Berlin, Germany) followed by biotin-labeled HLA B8-specific monoclonal antibody (One lambda Inc., La Jolla, Calif.). Following extensive washing with phosphate-buffered saline 1% FCS, cells were incubated with streptavidin-labeled red670 (Gibco BRL, Bethesda, Md.). Cells expressing both HLA B7 and HLA B8 were sorted twice and checked by fluorescence-activated cell sorting analysis for double-positive cells; 95 to 97% of the sorted cells were positive for both HLA markers. These chimeric cell lines were used as targets in a standard 51 Cr release assay. Recombinant vaccinia virus. A recombinant vaccinia virus construct encoding EBNA6 (Vacc.EBNA6) and a vaccinia virus construct made by insertion of the psc11 vector alone and negative for thymidine kinase (Vacc.TK ) have been previously described (12, 15). Cytotoxicity assay. CTL clones were screened in the standard 5-h chromium release assay (19). Briefly, 51 Cr-labeled LCLs or PHA blasts from B*4402, B*4403, and B*4405 individuals were incubated for 1 h without or with various concentrations of peptide epitope EENLLDVFRM. Following incubation, excess peptide was washed off with serum-free medium. Peptide EENLLDVFRM was synthesized by a simultaneous multiple peptide synthesis technique as described by Burrows et al. (4). CTL clones specific for peptide EENLLDVFRM were used as effectors in these assays. In some experiments, LCLs were infected with vaccinia virus constructs encoding EBNA6 at a multiplicity of infection of 10:1 for 12 to 14 h and used as targets in a standard chromium release assay. LDA. UM cells from HLA B*4402, B*4403, and B*4405 donors were distributed in graded numbers (twofold dilution) from to cells per well in round-bottom microtiter plates. Approximately irradiated (8,000 rads) autologous LCLs were added to give a total volume of 100 l. Twenty-four replicates were used at each concentration in each experiment, and an equal number of control wells were set up in which no stimulator cells were added. Cultures were fed on days 4 and 7 with 50 l of medium supplemented with 20 U of ril-2 and 30% (vol/vol) supernatant from MLA-144 cultures. On day 10, each CTL microculture was split into two replicates, and the cells were used as effectors in a standard 5-h 51 Cr release assay against autologous PHA blasts precoated with peptide EENLLDVFRM or left uncoated. Wells were scored as positive when the percent specific chromium release exceeded the mean release from control wells by 3 standard deviations. Limiting dilution analysis (LDA) was performed by the method of maximum likelihood estimation (7). Data from all experiments were compatible with the hypothesis of single-hit kinetics (P 0.4), and precursor estimates are given with 95% confidence limits. 1D-IEF. To analyze intracellular transport of HLA B44 subtypes, LCLs from donors CF (B*4402 ), CS (B*4403 ), and CM (B*4405) were starved for 30 min in 1 ml of methionine-cysteine-free RPMI 1640 supplemented with 10% FCS. The cells were pelleted and resuspended in 200 l of fresh methionine-cysteine-free medium and labeled with 150 Ci of [ 35 S]methioninecysteine for 10 min at 37 C. Cells were chased for 0, 5, 10, 30, 60, and 120 min in RPMI 1640 supplemented with 1 mm methionine-cysteine and 10% FCS and lysed in 1% digitonin-containing lysis mixture for 2 h on ice. After overnight preclearing with normal rabbit serum, the lysates were immunoprecipitated with W6/32 and the complexes were recovered with protein A-Sepharose. All samples were analyzed by one-dimensional isoelectric focusing (1D-IEF) (21). RESULTS Individuals carrying different subtypes of HLA B44 display hierarchy in their responses to the EBV EENLLDVFRM epitope. Preliminary studies with polyclonal EBV-specific CTLs from HLA B*4405 individuals consistently showed a very strong response to the EENLLDVFRM epitope, while a weak response was seen in HLA B*4402 and HLA B*4403 virus carriers (11a). To examine this issue at the clonal level, we compared the frequencies of CTL precursors (CTLp) specific for the EENLLDVFRM epitope in HLA B*4402, B*4403, and B*4405 individuals by LDA. EENLLDVFRM-specific CTLp were reactivated in vitro by stimulation of UM cells from donors CM (HLA B*4405 ), DM (HLA B*4405 ), CS (HLA B*4403 ), SE (HLA B*4403 ), BK (HLA B*4403 ), PM (HLA B*4403 ), CF (HLA B*4402 ), DY (HLA B*4402 ), IT (HLA B*4402 ), and KE (HLA B*4402 ) with autologous LCLs. Autologous PHA blasts precoated with peptide EENLLDVFRM were used as target cells in a chromium release assay. The representative data in Fig. 1 show that high frequencies of memory EENLLDVFRM-specific CTLs were detected in HLA B*4405 donors DM (1:3, ) and CM (1:7,592 1,754), while significantly lower numbers of CTLp specific for this epitope were seen in HLA B*4402 and B*4403 donors CS (74,377 30,103), BK (no CTLp detected), CF (141,885 88,407), IT (no CTLp detected), KE (40,760 15,100), and DY (no CTLp detected). Similarly, no EENLLDVFRM-specific CTLp were detected from donors SE and PM (data not shown). To exclude the possibility that competition among T-cell clones stimulated with LCLs, which recognize other EBV epitopes, was involved in restricting the CTL response to EENLLDVFRM in vitro, CTLp frequencies from donors CM, CS, and CF were also estimated in cultures stimulated with autologous UM cells precoated with synthetic peptide EENLLDVFRM. In accordance with the results shown in Fig. 1, the precursor frequency estimates were comparable to those detected following stimulation with autologous LCLs (data not shown). To further confirm the data obtained from LDA, EBVspecific CTL clones were established by using an agar cloning technique from HLA B*4405 (DM), B*4403 (CS and SE) and B*4402 (IT, KE, and DY) donors by stimulating UM cells with the autologous LCLs. As shown in Table 1, more than 80% of the EBV-specific clones from donor DM recognized EENLLDVFRM peptide. On the other hand, only 0 to 10% of the virus-specific CTL clones from donors CS, SE, IT, KE, and DY recognized this EBV epitope. A similar frequency of CTL clones specific for the EENLLDVFRM peptide was

3 VOL. 71, 1997 HLA SUBTYPE AND IMMUNODOMINANCE 7431 FIG. 1. CTLp frequencies for EBV epitope EENLLDVFRM in HLA B*4402, B*4403, and B*4405 individuals. Using LDA, we estimated the frequencies of CTLp for peptide EENLLDVFRM in UM cells from eight donors, HLA B*4402 (KE, DY, CF, and IT), HLA B*4403 (CS and BK), and HLA B*4405 (DM and CM). UM cells from these donors were stimulated with autologous LCLs as described in Materials and Methods. Reciprocal values of responder frequencies (f 1 ) are indicated. The shaded areas indicate 95% confidence limits. consistently seen in multiple blood samples from these donors over a period of 5 years. HLA B*4405 is more efficient in endogenous and exogenous presentation of the EENLLDVFRM epitope than HLA B*4402 and B*4403. To determine whether the disparity in CTLp frequency could be due to a difference in the presentation of the EENLLDVFRM epitope by different subtypes of HLA B44, we first compared the lysis of HLA B44 LCLs by CTL

4 7432 KHANNA ET AL. J. VIROL. TABLE 1. Frequency of EENLLDVFRM-specific CTL clones from HLA B44 individuals Donor B44 subtype Expt a No. of EBV-specific CTL clones b Total EENLLDVFRM specific DM B* (86.69 c ) (89.47) (82.75) (88.88) CS B* (5.88) (4.16) SE B* (0) (0) IT B* (0) (0) DY B* (0) KE B* (9.00) a Each of the experiment data presented above refers to a separate blood sample collected from these donors between 1990 and b EBV-specific CTL clones from HLA B44 donors were isolated as described in Materials and Methods. The EBV specificity of these CTL clones was confirmed as described previously (12). c Percentage of EBV-specific CTL clones specific for the EENLLDVFRM epitope. EENLLDVFRM specificity of the clone was confirmed as described elsewhere (4, 18). clones from HLA B*4405 (DM D10 and CM13) and B*4403 (CS27) donors at different effector-to-target (E/T) ratios. The data shown in Fig. 2a and b clearly demonstrate that B*4405 LCLs are more efficiently recognized than B*4402 and B*4403 LCLs. Surprisingly, a CTL clone from a B*4403 donor (CS27) also recognized B*4405 LCLs more efficiently than other HLA B44 LCLs (Fig. 2b). This disparity in the level of lysis was not due to a difference in the expression of EBNA6 gene in these cells, since comparable levels of EBNA6 were detected when analyzed by immunoblotting (data not shown). Moreover, a similar contrast differential in CTL recognition was also seen with another CTL clone (CM13) from a B*4405 donor CM (Fig. 2c). Interestingly, although overexpression of EBNA6 in B*4402 and B*4403 LCLs, following infection with a recombinant vaccinia virus encoding this antigen (Vacc.EBNA6), increased the level of CTL recognition by the CM13 clone, the differential pattern of lysis was maintained (Fig. 2c). In the next set of experiments, we examined the exogenous presentation efficiency of peptide EENLLDVFRM by the different subtypes of HLA B44. PHA blasts from donors DM (B*4405), CS (B*4403), and CF (B*4402) were presensitized with various concentrations of peptide EENLLDVFRM and then exposed to specific CTL clones or polyclonal CTLs. As illustrated in Fig. 3a to c, CTL clones from B*4405 (DM D10), B*4403 (CS27), and B*4402 (CF G12) donors recognized peptide-coated B*4405 PHA blasts more efficiently than either B*4402 or B*4403 target cells. The concentration of peptide EENLLDVFRM required for half-maximal lysis of B*4405 targets was approximately g/ml, while for B*4402 and B*4403 this concentration was 100-fold higher (0.1 to 0.3 g/ml). To confirm that this difference in half-maximal lysis was not restricted to these specific CTL clones, we repeated this analysis using an EENLLDVFRM-specific polyclonal CTL line from donor DM. As shown in Fig. 3d, the level of CTL recognition of peptide-sensitized B*4405 PHA blasts was again significantly more efficient compared to B*4402 and B*4403 PHA blasts. As a more stringent test to appraise the endogenous presentation efficiency of the HLA B*4405 subtype, B*4402 / FIG. 2. HLA B*4405 virus-infected cells are recognized more efficiently than B*4402 or B*4403 virus-infected cells. LCLs from B*4402, B*4403, and B*4405 donors were exposed to EENLLDVFRM-specific CTL clones (DM D10, CS27, and CM13) at different E/T ratios (a and b) or at a fixed E/T ratio of 4:1 (c). In some experiments, B44 LCLs were infected with recombinant vaccinia virus encoding EBNA6 antigen (which includes EENLLDVFRM epitope) and then used as targets in a standard 51 Cr release assay (c). B*4405 and B*4403 /B*4405 chimeric LCLs were generated, and the level of CTL lysis of these chimeric targets was compared to that of the B*4402, B*4403, and B*4405 LCLs. Data presented in Fig. 4 clearly demonstrate that although B*4402 and B*4403 LCLs were inefficiently recognized by the EENLLDVFRM-specific CTL clones DM D10, DM E7, and CS27, B*4402 /B*4405 and B*4403 /B*4405 chimeric LCLs were efficiently recognized by these CTL clones. The level of CTL lysis of these chimeric LCLs was comparable to that of B*4405 LCLs (Fig. 4).

5 VOL. 71, 1997 HLA SUBTYPE AND IMMUNODOMINANCE 7433 FIG. 3. CTL recognition of target cells expressing different subtypes of HLA B44 following exogenous sensitization with the EENLLDVFRM peptide. PHA blasts were sensitized with serial dilutions of each of the peptides and then exposed to either CTL clones CS27 (a), CF G12 (b), and DM D10 (c) or polyclonal CTL line DM EENL (d) at an E/T ratio of 4:1. Results are expressed as percent specific lysis. HLA B44 subtypes display significant variation in intracellular maturation and transport kinetics. Earlier studies have revealed that peptide loading of MHC molecules contributes directly to their intracellular transport and that many of the HLA class I alleles differ in efficiency of assembly (23, 27, 31). To explore the possibility that variation in CTL recognition of peptide bound to the various HLA B44 subtypes was also affected by the difference in the rate of maturation and intracellular transport of these subtypes, HLA B44 LCLs were studied by pulse-chase labeling and analyzed on 1D-IEF gels as described in Materials and Methods. Marked differences in the kinetics of assembly were noted between the different HLA class I alleles. Three representative gels showing pulse-chase analysis in HLA B*4405, B*4402, and B*4403 cell lines are illustrated in Fig. 5. One way of visualizing intracellular transport of newly assembled MHC class I molecules is by following the time-dependent disappearance of the nonsialylated form of class I heavy chains (indicating transport out of the ER lumen), which correlates with the appearance of the sialylated form upon arrival in the trans-golgi. In Fig. 5, it is clearly evident that the sialylated forms (labeled S) of the B*4405 allele can be seen at the 0-min chase, whereas the sialylated forms of B*4402 and B*4403 do not appear until after 5- and 30-min chases, respectively. Moreover, an analysis of unsialylated forms (labeled O) of the B44 alleles indicates a significant reduction in the 35 S-labeled B*4405 allele after 10 min of chase, while the unsialylated B*4403 and B*4402 alleles disappear after 30 min of chase. Thus, the B*4405 allele is assembled and transported more efficiently than the B*4402 and B*4403 alleles. DISCUSSION FIG. 4. HLA B*4402 /B*4405 and HLA B*4403 /B*4405 chimeric virusinfected cells are more efficiently recognized than B*4402 and B*4403 virusinfected cells. B*4405 LCLs were used as a positive control in this assay. Three different CTL clones, DM D10, DM E7, and CS27, were used. An E/T ratio of 2:1 was used in the assay. Results are expressed as percent specific lysis. Hierarchies among MHC class I-restricted CTL responses to intracellular pathogens have attracted considerable attention, since CTL epitopes which elicit strong responses might be preferential targets for immunotherapy and vaccine design. Very limited information is available on the factors responsible for these hierarchies in the CTL response. In the present study, we have investigated this issue in the context of a herpesvirus, EBV, that is known to establish a persistent infection in healthy individuals and is associated with a number of human malignancies. It is now well established that EBV-specific memory T cells are responsible for controlling the level of EBV-positive B lymphocytes and that the majority of these T

6 7434 KHANNA ET AL. J. VIROL. FIG. 5. Intracellular assembly and transport kinetics of HLA B44 subtypes. LCLs from donors B*4405 (A), B*4402 (B), and B*4403 (C) were bisynthetically labeled with 150 Ci of [ 35 S]methionine-cysteine for 10 min at 37 C. At the indicated time points, cells were lysed in digitonin-containing lysis mixture and then immunoprecipitated with W6/32. Complexes recovered with protein A-Sepharose were analyzed by 1D-IEF as described in Materials and Methods. O, unsialylated B44 molecules; S, sialylated B44 molecules. cells recognize epitopes included within the latent antigens of EBV (4, 13, 19). The CTL response to the HLA B44-restricted epitope EENLLDVFRM from EBNA6 in healthy individuals is a classic example of this phenomenon (3, 12). We have shown that healthy seropositive individuals carrying different subtypes of the HLA B44 antigen exhibit a discrete hierarchy in their CTL responses to the EENLLDVFRM epitope. This phenomenon appears to be strongly influenced by the peptide- MHC affinity and/or intracellular maturation and transport kinetics of MHC molecules. The possibility that the host MHC genotype can influence the strength of the CTL response to an individual CTL epitope has previously been proposed based on preliminary experiments with polyclonal CTL cultures from HLA B44 EBVimmune donors. In the present study, we have substantiated this assumption by demonstrating that CTLp specific for EENLLDVFRM are seen more frequently in B*4405 individuals than in B*4402 and B*4403 virus carriers. Furthermore, a similar pattern of response was also seen in the number of EENLLDVFRM-specific CTL clones isolated from B*4402, B*4403, and B*4405 donors over a period of 5 years. More than 60 to 80% of the CTL clones isolated from B*4405 individuals were EENLLDVFRM specific, while 0 to 5% of the CTL clones from B*4402 and B*4403 donors were specific for this epitope. To understand the molecular basis of this hierarchy, we analyzed the endogenous and exogenous presentation of EENLLDVFRM by target cells expressing different subtypes of HLA B44. In the first set of experiments, we noted that the specific CTL clones recognized B*4405 LCLs more efficiently than B*4402 or B*4403 EBV-infected cells. This phenomenon was seen not only with CTL clones from B*4405 individuals; surprisingly, CTLs specific for this epitope from a B*4403 donor also preferentially recognized B*4405 virus-infected cells. Similarly B*4405 PHA blasts, exogenously sensitized with EENLLDVFRM peptide, were consistently recognized more efficiently than B*4402 or B*4403 targets by all CTL clones irrespective of the HLA B44 subtype expressed by the donors from whom these effectors were isolated. Thus, it is unlikely that the difference in T-cell repertoire in these individuals is responsible for a strong or a weak response to EENLLDVFRM. On the other hand, the disparate CTL recognition of the EENLLDVFRM peptide in association with the different subtypes of HLA B44 likely reflects the relative abundance of corresponding class I- peptide complexes on target cells, and this may be influenced by two distinct factors. First, it is possible that the three subtypes of HLA B44 have different affinities for this peptide. Indeed, the data shown in Fig. 3 strongly argue in favor of this notion. Comparison of the amino acid sequences of the three different subtypes of HLA B44 shows that they differ by a single amino acid at position 116 or 156 (9, 35). It is possible that the difference between B*4405 and the other subtypes at position 116, which forms part of pocket F of the peptide binding cleft for MHC, influences the binding of EENLLDVFRM. Indeed, the tyrosine residue at position 116 of the B*4405 subtype is likely to interact more favorably with the methionine residue at the carboxy terminus of the peptide, while aspartic acid, a negatively charged residue, at the same position in B*4402 and B*4403 might reduce the efficiency of binding of the EENLLDVFRM peptide. It is possible that, in addition to differences in peptide binding efficiencies, the differential rates of intracellular assembly and transport of HLA B44 subtypes also contribute to the disparate efficiency in the presentation of the EENLLDVFRM epitope. Although earlier studies have shown that class I molecules display marked differences in the efficiency of assembly and transport (32), the biological significance of this observation is not understood. Data presented in this study argue in favor of a model in which efficient peptide binding and rapid intracellular assembly and transport kinetics by the MHC alleles contribute to a strong CTL response. It is important to mention here that since a single peptide is unlikely to induce an effect on the overall rate of transport of an individual class I allele, the specific rate in transport of each HLA B44 subtype is probably an intrinsic property of that allele. Thus, both efficient peptide binding and rapid transport kinetics have a synergistic effect on the presentation of the EENLLDVFRM epitope. Preliminary studies carried out in our laboratory have shown a similar strong correlation between other rapid-assembly alleles and immunodominant responses. HLA B8 and HLA B35 alleles, which are rapidly assembled and transported (32), consistently display strong responses, while weaker responses are seen for slowly assembling alleles such as HLA A3 and HLA B51 (unpublished observations). Taken together, the data presented in this study strongly suggest that class I alleles that efficiently bind peptides in the ER and are rapidly assembled and transported to the cell surface for presentation to CTL may provide a selective advantage in their ability to present peptide epitopes and confer a protective advantage against EBV infection. More importantly, these observations have significant implications for any future EBV vaccine designed to control acute or persistent viral infection. It is clear from the data presented here that HLA subtype distribution within different individuals might influence the long-term efficacy of CTL epitope-based vaccines.

7 VOL. 71, 1997 HLA SUBTYPE AND IMMUNODOMINANCE 7435 ACKNOWLEDGMENTS This work was supported by grants from the Queensland Cancer Fund, the National Health and Medical Research Council, and the National Cancer Institute (CA ). R.K. is an R. Douglas Wright Fellow. REFERENCES 1. Braciale, T. J., and V. L. Braciale Antigen presentation: structural themes and functional variations. Immunol. Today 12: Burrows, S. R., J. Gardner, R. Khanna, T. Steward, D. J. Moss, S. Rodda, and A. Suhrbier Five new cytotoxic T cell epitopes identified within Epstein-Barr virus nuclear antigen 3. J. Gen. Virol. 75: Burrows, S. R., I. S. Misko, T. B. Sculley, C. Schmidt, and D. J. Moss An Epstein-Barr virus-specific cytotoxic T-cell epitope present on A- and B-type transformants. J. Virol. 64: Burrows, S. R., S. J. Rodda, A. Suhrbier, H. M. Geysen, and D. J. Moss The specificity of recognition of a cytotoxic T lymphocyte epitope. Eur. J. Immunol. 22: Chen, W. S., S. Khilko, J. Fecondo, D. H. Margulies, and J. McCluskey Determinant selection of major histocompatibility complex class I- restricted antigenic peptides is explained by class I-peptide affinity and is strongly influenced by nondominant anchor residues. J. Exp. Med. 180: Daly, K., P. Nguyen, D. L. Woodland, and M. A. Blackman Immunodominance of major histocompatibility complex class I-restricted influenza virus epitopes can be influenced by the T-cell receptor repertoire. J. Virol. 69: Fazekas de St Groth, S The evaluation of limiting dilution assays. J. Immunol. Methods 49:R11 R Feltkamp, M. C., H. L. Smits, M. P. Vierboom, R. P. Minnaar, B. M. de Jongh, J. W. Drijfhout, J. ter Schegget, C. J. Melief, and W. M. Kast Vaccination with cytotoxic T lymphocyte epitope-containing peptide protects against a tumor induced by human papillomavirus type 16-transformed cells. Eur. J. Immunol. 23: Fleischhauer, K., D. Avila, F. Vilbois, C. Traversari, C. Bordignon, and H. J. Wallny Characterization of natural peptide ligands for HLA-B*4402 and -B*4403: implications for peptide involvement in allorecognition of a single amino acid change in the HLA-B44 heavy chain. Tissue Antigens 44: Jameson, S. C., and M. J. Bevan Dissection of major histocompatibility complex (MHC) and T cell receptor contact residues in a Kb-restricted ovalbumin peptide and an assessment of the predictive power of MHCbinding motifs. Eur. J. Immunol. 22: Kast, W. M., R. M. Brandt, J. Sidney, J. W. Drijfhout, R. T. Kubo, H. M. Grey, C. J. Melief, and A. Sette Role of HLA-A motifs in identification of potential CTL epitopes in human papillomavirus type 16 E6 and E7 proteins. J. Immunol. 152: a.Khanna, R., and S. R. Burrows. Unpublished data. 12. Khanna, R., S. R. Burrows, M. G. Kurilla, C. A. Jacob, I. S. Misko, T. B. Sculley, E. Kieff, and D. J. Moss Localization of Epstein-Barr virus cytotoxic T-cell epitopes using recombinant vaccinia: implications for vaccine development. J. Exp. Med. 176: Khanna, R., S. R. Burrows, and D. J. Moss Immune regulation in Epstein-Barr virus-associated diseases. Microbiol. Rev. 59: Khanna, R., C. A. Jacob, S. R. Burrows, M. G. Kurilla, E. Kieff, I. S. Misko, and D. J. Moss Expression of Epstein-Barr virus nuclear antigens in anti-igm-stimulated B cells following recombinant vaccinia infection and their recognition by human cytotoxic T-cells. Immunology 74: Khanna, R., C. A. Jacob, S. R. Burrows, and D. J. Moss Presentation of endogenous viral peptide epitopes by anti-cd40 stimulated human B cells following recombinant vaccinia infection. J. Immunol. Methods 164: Lipford, G. B., M. Hoffman, H. Wagner, and K. Heeg Primary in vivo responses to ovalbumin. Probing the predictive value of the Kb binding motif. J. Immunol. 150: Mamula, M. J The inability to process a self-peptide allows autoreactive T cells to escape tolerance. J. Exp. Med. 177: Moss, D. J., S. R. Burrows, G. D. Baxter, and M. F. Lavin T-cell-T cell killing is induced by specific epitopes: evidence for an apoptotic mechanism. J. Exp. Med. 173: Moss, D. J., I. S. Misko, S. R. Burrows, K. Burman, R. McCarthy, and T. B. Sculley Cytotoxic T-cell clones discriminate between A- and B-type Epstein-Barr virus transformants. Nature 331: Murray, R. J., M. G. Kurilla, J. M. Brooks, W. A. Thomas, M. Rowe, E. Kieff, and A. B. Rickinson Identification of target antigens for the human cytotoxic T-cell response to Epstein-Barr virus (EBV): implications for the immune control of EBV-positive malignancies. J. Exp. Med. 176: Neefjes, J. J., B. S. Breur Vriesendorp, G. A. Van Seventer, P. Ivanyi, and H. L. Ploegh An improved biochemical method for the analysis of HLA-class I antigens. Definition of new HLA-class I subtypes. Hum. Immunol. 16: Neefjes, J. J., F. Momburg, and G. J. Hammerling Selective and ATP-dependent translocation of peptides by the MHC-encoded transporter. Science 261: (Erratum, 264:16, 1994.) 23. Neefjes, J. J., and H. L. Ploegh Allele and locus-specific differences in cell surface expression and the association of HLA class I heavy chain with beta 2-microglobulin: differential effects of inhibition of glycosylation on class I subunit association. Eur. J. Immunol. 18: Neisig, A., J. Roelse, A. J. Sijts, F. Ossendorp, M. C. Feltkamp, W. M. Kast, C. J. Melief, and J. J. Neefjes Major differences in transporter associated with antigen presentation (TAP)-dependent translocation of MHC class I-presentable peptides and the effect of flanking sequences. J. Immunol. 154: Niedermann, G., S. Butz, H. G. Ihlenfeldt, R. Grimm, M. Lucchiari, H. Hoschutzky, G. Jung, B. Maier, and K. Eichmann Contribution of proteasome-mediated proteolysis to the hierarchy of epitopes presented by major histocompatibility complex class I molecules. Immunity 2: Oukka, M., N. Riche, and K. Kosmatopoulos A nonimmunodominant nucleoprotein-derived peptide is presented by influenza A virus-infected H-2b cells. J. Immunol. 152: Schumacher, T. N., M. T. Heemels, J. J. Neefjes, W. M. Kast, C. J. Melief, and H. L. Ploegh Direct binding of peptide to empty MHC class I molecules on intact cells and in vitro. Cell 62: Sheil, J. M., S. E. Shepherd, G. F. Klimo, and Y. Paterson Identification of an autologous insulin B chain peptide as a target antigen for H-2Kb-restricted cytotoxic T lymphocytes. J. Exp. Med. 175: Shepherd, J. C., T. N. Schumacher, P. G. Ashton-Rickardt, S. Imaeda, and H. L. Ploegh TAP1-dependent peptide translocation in vitro is ATP dependent and peptide selective. Cell 74: Sijts, A. J., F. Ossendorp, E. A. Mengede, P. J. van den Elsen, and C. J. Melief Immunodominant mink cell focus-inducing murine leukemia virus (MuLV)-encoded CTL epitope, identified by its MHC class I-binding motif, explains MuLV-type specificity of MCF-directed cytotoxic T lymphocytes. J. Immunol. 152: Townsend, A., C. Ohlën, J. Bastin, H. G. Ljunggren, L. Foster, and K. Kärre Association of class I major histocompatibility heavy and light chains induced by viral peptides. Nature 340: Townsend, A. R., J. Bastin, K. Gould, and G. G. Brownlee Cytotoxic T lymphocytes recognize influenza haemagglutinin that lacks a signal sequence. Nature 324: Wettstein, P. J., G. M. Van Bleek, and S. G. Nathenson Differential binding of a minor histocompatibility antigen peptide to H-2 class I molecules correlates with immune responsiveness. J. Immunol. 150: Wipke, B. T., S. C. Jameson, M. J. Bevan, and E. G. Pamer Variable binding affinities of listeriolysin O peptides for the H-2Kd class I molecule. Eur. J. Immunol. 23: Yao, Z., A. Volgger, S. Scholz, J. Bonisch, and E. D. Albert Nucleotide sequence of a novel HLA-B44 subtype B*4405. Immunogenetics 40:310.